Thermally Activated Gold‐Mediated Transition Metal Dichalcogenide Exfoliation and a Unique Gold‐Mediated Transfer

Layered materials like transition metal dichalcogenides (TMDCs) enable exciting new physics in their 2D limit. Combined with successful demonstrations of 2D transistors and devices, the need for high‐quality large‐scale monolayers increases. In this light, scalable gold‐mediated exfoliations attract broad attention to supersede the traditional scotch tape method as a means for high‐quality materials. Gold proved to be suitably adhesive for the exfoliation of several 2D materials, including TMDCs. Previously reported methods rely on a simple press and peel mechanism. However, herein, a gold‐mediated exfoliation enabled by low‐temperature annealing is presented for the first time. This simple modification potentially increases the range of external conditions under which gold‐mediated exfoliations operate in a robust manner. The exfoliation achieves scaling with parent crystal areas, rendering it on par with previously reported methods. On top of that, a unique gold‐mediated transfer concept is introduced, where gold is repurposed as a metallic (polymer‐free) transfer membrane. The transfer allows the deterministic and clean relocation of the exfoliated monolayers onto technologically relevant substrates like SiO2/Si. The process is benchmarked using MoS2 as the prototypical TMDC and monolayer areas up to ≈80 mm2 are successfully exfoliated and transferred.

Layered materials like transition metal dichalcogenides (TMDCs) enable exciting new physics in their 2D limit. Combined with successful demonstrations of 2D transistors and devices, the need for high-quality large-scale monolayers increases. In this light, scalable gold-mediated exfoliations attract broad attention to supersede the traditional scotch tape method as a means for high-quality materials. Gold proved to be suitably adhesive for the exfoliation of several 2D materials, including TMDCs. Previously reported methods rely on a simple press and peel mechanism. However, herein, a gold-mediated exfoliation enabled by low-temperature annealing is presented for the first time. This simple modification potentially increases the range of external conditions under which goldmediated exfoliations operate in a robust manner. The exfoliation achieves scaling with parent crystal areas, rendering it on par with previously reported methods. On top of that, a unique gold-mediated transfer concept is introduced, where gold is repurposed as a metallic (polymer-free) transfer membrane. The transfer allows the deterministic and clean relocation of the exfoliated monolayers onto technologically relevant substrates like SiO 2 /Si. The process is benchmarked using MoS 2 as the prototypical TMDC and monolayer areas up to %80 mm 2 are successfully exfoliated and transferred.

RAPID RESEARCH LETTER
www.pss-rapid.com Phys. Status Solidi RRL 2020, 14,2000408 Supporting Information), [11][12][13][14] rendering MoS 2 a perfect TMDC prototype for benchmarking gold-assisted exfoliations. Smooth and clean gold surfaces provide the Au/TMDC vdW interactions necessary to overcome the TMDC interlayer interactions, leading to exfoliation of the interfacing layer. [11] The template-strip method provides these gold surfaces by cleaving an gold film off an smooth template substrate, accessing the uncontaminated part of the former gold template interface. [15] In this work, template-stripped gold substrates acted as ideal exfoliation substrates to yield large-area and highly continuous MoS 2 . An unprecedent thermally activated exfoliation mechanism was observed. The process was complemented by reusing gold as the transfer membrane, thereby implementing a novel polymer-free transfer approach.
The developed TMDC exfoliation process is shown in Figure 1a with MoS 2 as the test bed. The metal-stacks can be peeled off the template wafer (template strip), allowing for wafer-scale batches of ultraflat gold exfoliation substrates (see details in Figure S1, Supporting Information). The exfoliation starts with revealing fresh surfaces of MoS 2 and gold, combining them on a hotplate, and finishes by peeling off the MoS 2 parent crystal. This leads to ultra-large MoS 2 layers exfoliated onto gold, as shown in the optical micrograph in Figure 1b. The MoS 2 single-layer areas (1 L) are indicated and measure up to 37 mm 2 in this example. Some multilayer areas are apparent, but the single-layer coverage outweighs the multilayer fraction by far, supporting the notion of the successful exfoliation of MoS 2 layers in conformal contact with gold. Figure 1c shows a zoom-in of the monolayer region of Figure 1b, revealing the monolayer's continuity over several square millimeters. The layered nature is confirmed using atomic force microscopy (AFM), as shown in Figure 1 d,e. The layer's apparent step height of % 0.6 nm is consistent with reported values. [16] No photoluminescence (PL) was observed for MoS 2 on gold, due to quenching by the metal. [11,17] Surprisingly, exfoliation did not work without the annealing step at 200 C. This is distinct from the previously reported gold-based exfoliations, [11][12][13][14]18] where no annealing was needed. Following previous reports one would expect room-temperature exfoliation at the smooth template-stripped Au surface (RMS %0.3 nm, Figure S1, Supporting Information), driven by strong vdW interactions at the interface, as described by Velický et al. for smooth and clean gold (RMS %0.7 nm) [11] in conformal contact with TMDCs. However, we were not able to exfoliate at room temperature. Either the interface is already contaminated or the gold growth at the SiO 2 template changes the crystallinity of the resulting gold surface (for nontemplated evaporation mostly Au(111) grains are expected [11] ). Therefore, thermally assisted "self-cleansing" mechanisms [7] or surface reconstructions [7] at the gold/MoS 2 interface (effectively a 3D-2D vdW heterostructure) might be possible explanations. The process also worked at lower temperatures down to 150 C, yet reproducibility and yield decreased. Cleaving the tape off the hot sample, skipping the cooling step, did not change the exfoliation results. The tape was peeled slowly (% 0.5-1 mm s À1 ) by hand with an approximate peel angle between 60 and 90 . The details of the (last) peeling step did not show a notable difference in the exfoliation yield. However, the initial peeling of the parent crystal might be more sensitive to such details as ripples and tear-ins could be introduced on the otherwise smooth crystal surface, potentially decreasing exfoliation yield and increasing multilayers exfoliated from teared parent crystal sites. A gentle handling of the parent crystal therefore seems advisable. With 200 C as the ideal temperature the exfoliation yields highly continuous monolayers, potentially allowing increased reproducibility overall due to the reduced influence of external conditions like carbon contaminants on gold. [11] Scaling of the exfoliation process was investigated by comparing the parent crystal surface areas in contact with gold with the areas of exfoliated monolayers (Figure 2a  www.advancedsciencenews.com www.pss-rapid.com exfoliation (Figure 2c). The exfoliation coverage clearly reproduced the wrinkled pattern of the parent crystal, following the limits of conformal contact between parent crystal and gold. The highlighted feature (white dotted contour) depicts a fold in the parent crystal, translating to a nonexfoliated area. With the parent crystal as the limiting factor, we realized a scalable method, and using larger (and smoother) crystals led to the anticipated upscaling of the exfoliated monolayer area. Utilizing the largest MoS 2 crystal to our disposal (% 115 mm 2 surface area), 82 mm 2 of monolayer area was exfoliated (Figure 2d). The herein reported monolayer area presents one of the largest for any gold-mediated exfoliation process to date (see Table S1, Supporting Information), yet the compelling argument is the achieved scaling in a robust and accessible method. A gold-mediated transfer was also implemented utilizing the copper layer. The introduced copper layer can be etched orthogonally to gold using FeCl 3 , leaving MoS 2 on a floating gold foil acting as a metal (polymer-free) transfer membrane. The polymer-free transfer process is schematically shown in Figure 3a. The foil on the target substrate is dried in air, annealed to promote conformal contact between the substrate and MoS 2 , and then etched using aqueous KI/I 2 to release the MoS 2 monolayer. The KI/I 2 etchant does not exhibit any MoS 2 etching (FeCl 3 is at no point in contact with MoS 2 ). The transfer process is capable of deterministically transferring the complete monolayer area onto SiO 2 , as shown in Figure 3b. The MoS 2 1L area (82 mm 2 ) is conserved, as well as the relative positions of the MoS 2 features. AFM reveals a SiO 2 /MoS 2 -1L step height of % 0.9 nm, consistent with previous reports, [19] and proves the wrinkle-free transfer of the monolayer (Figure 3c). The optical characterization shown in Figure 3 d,e features PL emission [17,20,21] for single-layer MoS 2 and the characteristic E 1 2g and A 1g Raman mode frequency, with A 1g -E 1 2g equating to 18.8 cm À1 , as expected for single-layer MoS 2 . [22,23] These results indicate intact monolayers after exfoliation and transfer. The novel polymer-free transfer unlocks the monolayers for use on any substrate. Several polymer-based transfer processes were also tested, which proved to be reliable as well, yet plagued by ubiquitous polymeric residues ( Figure S2, Supporting Information). The large monolayer areas achieved here allow to study the electronic structure of MoS 2 , in principle on any substrate, using standard photoemission spectroscopy without experiencing issues with sample charging [17] (details in Figure S4, Supporting Information). The lack of Mo 6þ peaks in the Mo3d core-level region indicates the absence of oxidation, [24][25][26] rendering our exfoliation/transfer process a suitable route for high-quality monolayers. Furthermore, the complete etching and removal of gold after transfer were confirmed ( Figure S5, Supporting Information). To probe the monolayer quality, ultralarge field-effect transistors (1 mm Â 30 μm for W Â L) on SiO 2 were fabricated with exfoliated MoS 2 monolayers ( Figure S6). The devices showed decent performance (room-temperature mobility μ ¼ 8.9 cm 2 V À1 s À1 , I on /I off ¼ 1 Â 10 5 ), a statement of the intact semiconducting properties and the material quality over a large area. In summary, we present a scalable thermally activated gold-mediated TMDC exfoliation and a novel gold-mediated transfer. Resulting monolayers show high continuity and were proved to maintain the high starting material quality over large areas. The template-strip method paired with silicon wafers allowed us to produce wafer-scale batches of exfoliation substrates. These gold substrates are easily stored and are always just one strip away to initiate ultralarge exfoliations. The herein disclosed process may boost the technological exploitation of 2D TMDC and their heterostructures fuelled by a robust high-throughput exfoliation process with potential for automation. Ultimately, scaling with initial parent crystal areas hints at the importance of large and smooth crystals for scaling up exfoliations into technologically relevant realms.

Experimental Section
Gold Exfoliation Substrates: Polished silicon wafers (Siegert Wafer, <100>, 525 μm thickness) served as ultraflat templates. The wafers were sonicated for 5 min each in acetone, isopropyl alcohol (IPA), and deionized (DI) water and then dried with a nitrogen gun. A 5 min O 2 -plasma treatment (Diener Femto PCCE, 0.35 mbar, 500 W, 40 kHz) followed by DI water rinse and 5 min DI water sonication followed. Around 200 nm Au www.advancedsciencenews.com www.pss-rapid.com was deposited via physical vapor deposition (%1.0 Å s À1 , %10 À7 mbar) followed by 800 nm copper (%2.5 Å s À1 ). Glass slides (1.5 Â 1.0 Â 0.7 mm) were glued onto the metal-coated wafer (Ossila Encapsulation Epoxy S132) and UV cured (254 and 365 nm, 1.5 h). Gold-Assisted Exfoliation: MoS 2 (2D semiconductors, synthetic MoS 2 crystal) was cleaved with heat-resistant Kapton tape and pressed onto a freshly stripped Au substrate. The stack was annealed on a hotplate in ambient conditions at 200 C for 60 s and taken from the hotplate to cool down for 10-20 s before peeling the tape.
Gold-Assisted Transfer: Exfoliated MoS 2 on Au was floated on aqueous FeCl 3 (33 wt%). The sample was left in the etchant until the gold foil floating freely on top (%12 h). The foil was fished with a clean wafer piece and transferred into a beaker with DI water to clean off etchant residues. The foil was transferred into a fresh DI water beaker where the surface tension was lowered with IPA to sink and flip the foil. The flipped foil was transferred into a beaker containing the solvent applicable for the target substrate. The substrate was positioned below the foil and the solvent was drained with a pipette, transferring the foil onto the substrate. The sample was dried in air for 1 h before baking on a hotplate (80 C 1 h, 150 C 30 min). Au was etched for 30 min with aqueous KI/I 2 etchant (Sigma Aldrich, "Au etchant, standard") which was deposited on top of the Au foil with a pipette. To wash off Au etchant residues, the substrate was washed with acetone, followed by immersion in acetone and then ethanol. The last solvent in the transfer should wet the target substrate well to allow even spreading of the foil while not exhibiting high vapor pressures at room temperature to avoid blistering of the gold foil during drying.
AFM Measurements: AFM was conducted in ambient conditions with a Bruker Dimension Icon using PeakForce Tapping with a ScanAsyst-Air tip (Bruker).
Raman and PL Measurements: Raman and PL spectroscopy were conducted using a confocal microscope setup (XploRA, Horiba Ltd.) with a 532 nm laser excitation source and 100x objective (%1 μm laser spot size). Laser power was kept below 0.14 mW. All measurements were carried out in ambient conditions. Photoelectron Spectroscopy: Experiments were conducted at the ENERGIZE end station at Bessy II, using the Mg anode of a Scienta Omicron DAR400 X-ray source (1254.6 eV) and a Focus HIS 13 helium discharge lamp (21.2 eV) for excitation. The kinetic energy of the emitted electrons was measured by a Scienta Omicron DA30 analyzer and the sample was biased with À10 V when measuring the secondary electron cutoff. Before measurements the sample was annealed at 300 C in ultra-high vacuum (base pressure > 10 À9 mbar) for several hours.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.